Face Generation¶

In this project, you'll use generative adversarial networks to generate new images of faces.

Get the Data¶

You'll be using two datasets in this project:

MNIST
CelebA

Since the celebA dataset is complex and you're doing GANs in a project for the first time, we want you to test your neural network on MNIST before CelebA. Running the GANs on MNIST will allow you to see how well your model trains sooner.

If you're using FloydHub, set data_dir to "/input" and use the FloydHub data ID "R5KrjnANiKVhLWAkpXhNBe".

data_dir = './data'

# FloydHub - Use with data ID "R5KrjnANiKVhLWAkpXhNBe"
#data_dir = '/input'


"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import helper

helper.download_extract('mnist', data_dir)
helper.download_extract('celeba', data_dir)

Found mnist Data
Found celeba Data

Explore the Data¶

MNIST¶

As you're aware, the MNIST dataset contains images of handwritten digits. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
%matplotlib inline
import os
from glob import glob
from matplotlib import pyplot

mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')
pyplot.imshow(helper.images_square_grid(mnist_images, 'L'), cmap='gray')

<matplotlib.image.AxesImage at 0x7fe1698682e8>

CelebA¶

The CelebFaces Attributes Dataset (CelebA) dataset contains over 200,000 celebrity images with annotations. Since you're going to be generating faces, you won't need the annotations. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28, 28, 'RGB')
pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))

<matplotlib.image.AxesImage at 0x7fe169821630>

Preprocess the Data¶

Since the project's main focus is on building the GANs, we'll preprocess the data for you. The values of the MNIST and CelebA dataset will be in the range of -0.5 to 0.5 of 28x28 dimensional images. The CelebA images will be cropped to remove parts of the image that don't include a face, then resized down to 28x28.

The MNIST images are black and white images with a single color channel while the CelebA images have 3 color channels (RGB color channel).

Build the Neural Network¶

You'll build the components necessary to build a GANs by implementing the following functions below:

model_inputs
discriminator
generator
model_loss
model_opt
train

Check the Version of TensorFlow and Access to GPU¶

This will check to make sure you have the correct version of TensorFlow and access to a GPU

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
from distutils.version import LooseVersion
import warnings
import tensorflow as tf

# Check TensorFlow Version
assert LooseVersion(tf.__version__) >= LooseVersion('1.0'), 'Please use TensorFlow version 1.0 or newer.  You are using {}'.format(tf.__version__)
print('TensorFlow Version: {}'.format(tf.__version__))

# Check for a GPU
if not tf.test.gpu_device_name():
    warnings.warn('No GPU found. Please use a GPU to train your neural network.')
else:
    print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))

TensorFlow Version: 1.0.1
Default GPU Device: /gpu:0

Input¶

Implement the model_inputs function to create TF Placeholders for the Neural Network. It should create the following placeholders:

Real input images placeholder with rank 4 using image_width, image_height, and image_channels.
Z input placeholder with rank 2 using z_dim.
Learning rate placeholder with rank 0.

Return the placeholders in the following the tuple (tensor of real input images, tensor of z data)

import problem_unittests as tests

def model_inputs(image_width, image_height, image_channels, z_dim):
    """
    Create the model inputs
    :param image_width: The input image width
    :param image_height: The input image height
    :param image_channels: The number of image channels
    :param z_dim: The dimension of Z
    :return: Tuple of (tensor of real input images, tensor of z data, learning rate)
    """
    # TODO: Implement Function
    input_real = tf.placeholder(tf.float32, (None, image_width, image_height, image_channels), name='input_real')
    input_z = tf.placeholder(tf.float32, (None, z_dim), name='input_z')
    learning_rate = tf.placeholder(tf.float32, name='learning_rate')
    

    return input_real, input_z, learning_rate


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_inputs(model_inputs)

Tests Passed

Discriminator¶

Implement discriminator to create a discriminator neural network that discriminates on images. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "discriminator" to allow the variables to be reused. The function should return a tuple of (tensor output of the discriminator, tensor logits of the discriminator).

def discriminator(images, reuse=False):
    """
    Create the discriminator network
    :param images: Tensor of input image(s)
    :param reuse: Boolean if the weights should be reused
    :return: Tuple of (tensor output of the discriminator, tensor logits of the discriminator)
    """
    # TODO: Implement Function
    alpha = 0.2
    with tf.variable_scope('discriminator', reuse=reuse):
        # Input layer is 28x28x3
        x1 = tf.layers.conv2d(images, 64, 5, strides=2, padding='same')
        relu1 = tf.maximum(alpha * x1, x1)
        # 14x14x64
        
        x2 = tf.layers.conv2d(relu1, 128, 5, strides=2, padding='same')
        bn2 = tf.layers.batch_normalization(x2, training=True)
        relu2 = tf.maximum(alpha * bn2, bn2)
        # 7x7x128
        
        x3 = tf.layers.conv2d(relu2, 256, 5, strides=2, padding='same')
        bn3 = tf.layers.batch_normalization(x3, training=True)
        relu3 = tf.maximum(alpha * bn3, bn3)
        # 4x4x256

        # Flatten it
        flat = tf.reshape(relu3, (-1, 4*4*256))
        logits = tf.layers.dense(flat, 1)
        out = tf.sigmoid(logits)
        
        return out, logits


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_discriminator(discriminator, tf)

Tests Passed

Generator¶

Implement generator to generate an image using z. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "generator" to allow the variables to be reused. The function should return the generated 28 x 28 x out_channel_dim images.

def generator(z, out_channel_dim, is_train=True):
    """
    Create the generator network
    :param z: Input z
    :param out_channel_dim: The number of channels in the output image
    :param is_train: Boolean if generator is being used for training
    :return: The tensor output of the generator
    """
    # TODO: Implement Function
    alpha = 0.2
    reuse = not is_train
    with tf.variable_scope('generator', reuse=reuse):
        # First fully connected layer
        x1 = tf.layers.dense(z, 4*4*512)
        # Reshape it to start the convolutional stack
        x1 = tf.reshape(x1, (-1, 4, 4, 512))
        bn1 = tf.layers.batch_normalization(x1, training=is_train)
        lrelu1 = tf.maximum(alpha * bn1, bn1)
        # 4x4x512 now
        
        upsample2 = tf.image.resize_nearest_neighbor(lrelu1, (7,7))
        conv2 = tf.layers.conv2d(upsample2, 256, 5, padding='same', activation=None)
        # x2 = tf.layers.conv2d_transpose(x1, 256, 5, strides=2, padding='same')
        bn2 = tf.layers.batch_normalization(conv2, training=is_train)
        lrelu2 = tf.maximum(alpha * bn2, bn2)
        # 7x7x256 now
        
        upsample3 = tf.image.resize_nearest_neighbor(lrelu2, (14,14))
        conv3 = tf.layers.conv2d(upsample3, 128, 5, padding='same', activation=None)
        bn3 = tf.layers.batch_normalization(conv3, training=is_train)
        lrelu3 = tf.maximum(alpha * bn3, bn3)
        # 16x16x128 now
        
        # Output layer
        upsample3 = tf.image.resize_nearest_neighbor(lrelu3, (28,28))
        logits = tf.layers.conv2d(upsample3, out_channel_dim, 5, padding='same')
        # 32x32x3 now
        
        out = tf.tanh(logits)
        
        return out


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_generator(generator, tf)

Tests Passed

Loss¶

Implement model_loss to build the GANs for training and calculate the loss. The function should return a tuple of (discriminator loss, generator loss). Use the following functions you implemented:

discriminator(images, reuse=False)
generator(z, out_channel_dim, is_train=True)

def model_loss(input_real, input_z, out_channel_dim):
    """
    Get the loss for the discriminator and generator
    :param input_real: Images from the real dataset
    :param input_z: Z input
    :param out_channel_dim: The number of channels in the output image
    :return: A tuple of (discriminator loss, generator loss)
    """
    # TODO: Implement Function
    gen_model = generator(input_z,out_channel_dim,is_train=True)
    d_model_real,d_logits_real =discriminator(input_real,reuse=False)
    d_model_fake,d_logits_fake =discriminator(gen_model,reuse=True)
    
    d_loss_real = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_real, labels = tf.ones_like(d_model_real) * 0.9))
    d_loss_fake = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels = tf.zeros_like(d_model_fake)))
    d_loss = d_loss_real + d_loss_fake
    
    g_loss = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels = tf.ones_like(d_model_fake)))

    
    return d_loss, g_loss


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_loss(model_loss)

Tests Passed

Optimization¶

Implement model_opt to create the optimization operations for the GANs. Use tf.trainable_variables to get all the trainable variables. Filter the variables with names that are in the discriminator and generator scope names. The function should return a tuple of (discriminator training operation, generator training operation).

def model_opt(d_loss, g_loss, learning_rate, beta1):
    """
    Get optimization operations
    :param d_loss: Discriminator loss Tensor
    :param g_loss: Generator loss Tensor
    :param learning_rate: Learning Rate Placeholder
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :return: A tuple of (discriminator training operation, generator training operation)
    """
    # TODO: Implement Function
    t_vars = tf.trainable_variables()
    d_vars = [var for var in t_vars if var.name.startswith('discriminator')]
    g_vars = [var for var in t_vars if var.name.startswith('generator')]
   
    with tf.control_dependencies(tf.get_collection(tf.GraphKeys.UPDATE_OPS)):
        g_opt_train = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(g_loss, var_list=g_vars)
        d_opt_train = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(d_loss, var_list=d_vars)
    
        return d_opt_train, g_opt_train


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_opt(model_opt, tf)

Tests Passed

Neural Network Training¶

Show Output¶

Use this function to show the current output of the generator during training. It will help you determine how well the GANs is training.

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import numpy as np

def show_generator_output(sess, n_images, input_z, out_channel_dim, image_mode):
    """
    Show example output for the generator
    :param sess: TensorFlow session
    :param n_images: Number of Images to display
    :param input_z: Input Z Tensor
    :param out_channel_dim: The number of channels in the output image
    :param image_mode: The mode to use for images ("RGB" or "L")
    """
    cmap = None if image_mode == 'RGB' else 'gray'
    z_dim = input_z.get_shape().as_list()[-1]
    example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])

    samples = sess.run(
        generator(input_z, out_channel_dim, False),
        feed_dict={input_z: example_z})

    images_grid = helper.images_square_grid(samples, image_mode)
    pyplot.imshow(images_grid, cmap=cmap)
    pyplot.show()

Train¶

Implement train to build and train the GANs. Use the following functions you implemented:

model_inputs(image_width, image_height, image_channels, z_dim)
model_loss(input_real, input_z, out_channel_dim)
model_opt(d_loss, g_loss, learning_rate, beta1)

Use the show_generator_output to show generator output while you train. Running show_generator_output for every batch will drastically increase training time and increase the size of the notebook. It's recommended to print the generator output every 100 batches.

def train(epoch_count, batch_size, z_dim, learning_rate, beta1, get_batches, data_shape, data_image_mode):
    """
    Train the GAN
    :param epoch_count: Number of epochs
    :param batch_size: Batch Size
    :param z_dim: Z dimension
    :param learning_rate: Learning Rate
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :param get_batches: Function to get batches
    :param data_shape: Shape of the data
    :param data_image_mode: The image mode to use for images ("RGB" or "L")
    """
    # TODO: Build Model
    
    losses = []
    steps = 0
    
    _, image_width, image_height, image_channels = data_shape
    input_real, input_z , learning_rate_pl = model_inputs(image_width,image_height,image_channels,z_dim)
    
    d_loss, g_loss = model_loss(input_real, input_z, image_channels)
    d_opt, g_opt = model_opt(d_loss, g_loss,learning_rate_pl, beta1)
    n_images = 25
    
    with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        for epoch_i in range(epoch_count): 
            for batch_images in get_batches(batch_size):
                # TODO: Train Model
                steps += 1
                
                # Sample random noise for G
                batch_z = np.random.uniform(-1, 1, size=(batch_size, z_dim))
                batch_images = batch_images * 2.0
                
                _ = sess.run(d_opt, feed_dict={input_real:batch_images,input_z:batch_z,learning_rate_pl:learning_rate})
                _ = sess.run(g_opt, feed_dict={input_real:batch_images,input_z:batch_z,learning_rate_pl:learning_rate})
                # sess.run(g_opt, feed_dict={input_real:batch_images,input_z:batch_z,learning_rate_pl:learning_rate})
                
                if steps % 10 == 0 :
                    train_loss_d = d_loss.eval({input_real:batch_images,input_z:batch_z,learning_rate_pl:learning_rate})
                    train_loss_g = g_loss.eval({input_real:batch_images,input_z:batch_z,learning_rate_pl:learning_rate})
                    print("Epoch {}/{}...".format(epoch_i+1, epoch_count),
                      "Discriminator Loss: {:.4f}...".format(train_loss_d),
                      "Generator Loss: {:.4f}".format(train_loss_g))
                
                    losses.append((train_loss_d, train_loss_g))
                    
                if steps % 100 == 0:
                    show_generator_output(sess,n_images,input_z,image_channels,data_image_mode)

MNIST¶

Test your GANs architecture on MNIST. After 2 epochs, the GANs should be able to generate images that look like handwritten digits. Make sure the loss of the generator is lower than the loss of the discriminator or close to 0.

batch_size = 128
z_dim = 512
learning_rate = 0.0002
beta1 = 0.5


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 25

mnist_dataset = helper.Dataset('mnist', glob(os.path.join(data_dir, 'mnist/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, mnist_dataset.get_batches,
          mnist_dataset.shape, mnist_dataset.image_mode)

Epoch 1/25... Discriminator Loss: 0.4619... Generator Loss: 2.6462
Epoch 1/25... Discriminator Loss: 0.3906... Generator Loss: 3.9583
Epoch 1/25... Discriminator Loss: 0.3677... Generator Loss: 3.8449
Epoch 1/25... Discriminator Loss: 0.7656... Generator Loss: 3.0281
Epoch 1/25... Discriminator Loss: 1.0453... Generator Loss: 4.5764
Epoch 1/25... Discriminator Loss: 0.6831... Generator Loss: 2.8820
Epoch 1/25... Discriminator Loss: 0.5185... Generator Loss: 3.2898
Epoch 1/25... Discriminator Loss: 0.4767... Generator Loss: 3.0170
Epoch 1/25... Discriminator Loss: 0.5933... Generator Loss: 2.5292
Epoch 1/25... Discriminator Loss: 0.6219... Generator Loss: 2.2008

Epoch 1/25... Discriminator Loss: 1.1004... Generator Loss: 1.1556
Epoch 1/25... Discriminator Loss: 0.6982... Generator Loss: 1.9497
Epoch 1/25... Discriminator Loss: 0.8090... Generator Loss: 1.5425
Epoch 1/25... Discriminator Loss: 1.2682... Generator Loss: 0.6667
Epoch 1/25... Discriminator Loss: 0.9897... Generator Loss: 1.4328
Epoch 1/25... Discriminator Loss: 2.1235... Generator Loss: 3.2972
Epoch 1/25... Discriminator Loss: 1.2764... Generator Loss: 1.3099
Epoch 1/25... Discriminator Loss: 1.0558... Generator Loss: 1.0448
Epoch 1/25... Discriminator Loss: 1.1302... Generator Loss: 1.0799
Epoch 1/25... Discriminator Loss: 1.0854... Generator Loss: 1.0294

Epoch 1/25... Discriminator Loss: 1.1058... Generator Loss: 1.0616
Epoch 1/25... Discriminator Loss: 1.1063... Generator Loss: 1.0626
Epoch 1/25... Discriminator Loss: 1.0034... Generator Loss: 1.2190
Epoch 1/25... Discriminator Loss: 1.0519... Generator Loss: 1.1118
Epoch 1/25... Discriminator Loss: 1.2629... Generator Loss: 0.7075
Epoch 1/25... Discriminator Loss: 0.9590... Generator Loss: 1.7456
Epoch 1/25... Discriminator Loss: 1.1463... Generator Loss: 0.8991
Epoch 1/25... Discriminator Loss: 1.0308... Generator Loss: 1.3357
Epoch 1/25... Discriminator Loss: 0.9856... Generator Loss: 1.1072
Epoch 1/25... Discriminator Loss: 1.0220... Generator Loss: 1.3656

Epoch 1/25... Discriminator Loss: 1.3399... Generator Loss: 2.4100
Epoch 1/25... Discriminator Loss: 1.2802... Generator Loss: 0.6681
Epoch 1/25... Discriminator Loss: 1.0905... Generator Loss: 1.3547
Epoch 1/25... Discriminator Loss: 1.1189... Generator Loss: 0.8986
Epoch 1/25... Discriminator Loss: 0.9929... Generator Loss: 1.3809
Epoch 1/25... Discriminator Loss: 1.3192... Generator Loss: 0.6472
Epoch 1/25... Discriminator Loss: 1.0704... Generator Loss: 1.0200
Epoch 1/25... Discriminator Loss: 1.1015... Generator Loss: 0.8406
Epoch 1/25... Discriminator Loss: 1.2030... Generator Loss: 0.7086
Epoch 1/25... Discriminator Loss: 1.0944... Generator Loss: 0.9246

Epoch 1/25... Discriminator Loss: 1.2395... Generator Loss: 1.4584
Epoch 1/25... Discriminator Loss: 1.1169... Generator Loss: 1.1963
Epoch 1/25... Discriminator Loss: 1.3102... Generator Loss: 0.6116
Epoch 1/25... Discriminator Loss: 1.1928... Generator Loss: 0.9928
Epoch 1/25... Discriminator Loss: 1.3388... Generator Loss: 0.5786
Epoch 1/25... Discriminator Loss: 1.3170... Generator Loss: 0.6986
Epoch 2/25... Discriminator Loss: 1.4323... Generator Loss: 0.5025
Epoch 2/25... Discriminator Loss: 1.1575... Generator Loss: 0.8527
Epoch 2/25... Discriminator Loss: 1.2025... Generator Loss: 0.7869
Epoch 2/25... Discriminator Loss: 1.0806... Generator Loss: 1.1057

Epoch 2/25... Discriminator Loss: 1.1231... Generator Loss: 0.9940
Epoch 2/25... Discriminator Loss: 1.3860... Generator Loss: 0.5039
Epoch 2/25... Discriminator Loss: 1.3572... Generator Loss: 0.5608
Epoch 2/25... Discriminator Loss: 1.1305... Generator Loss: 1.6062
Epoch 2/25... Discriminator Loss: 1.1491... Generator Loss: 0.9504
Epoch 2/25... Discriminator Loss: 1.3015... Generator Loss: 0.6015
Epoch 2/25... Discriminator Loss: 1.1096... Generator Loss: 1.6594
Epoch 2/25... Discriminator Loss: 1.2712... Generator Loss: 0.6233
Epoch 2/25... Discriminator Loss: 1.1372... Generator Loss: 1.1727
Epoch 2/25... Discriminator Loss: 1.1219... Generator Loss: 0.8192

Epoch 2/25... Discriminator Loss: 1.1778... Generator Loss: 0.7424
Epoch 2/25... Discriminator Loss: 1.2164... Generator Loss: 0.6980
Epoch 2/25... Discriminator Loss: 1.2170... Generator Loss: 0.7748
Epoch 2/25... Discriminator Loss: 1.4209... Generator Loss: 0.5858
Epoch 2/25... Discriminator Loss: 1.4282... Generator Loss: 0.5061
Epoch 2/25... Discriminator Loss: 1.2264... Generator Loss: 0.7017
Epoch 2/25... Discriminator Loss: 1.4024... Generator Loss: 0.5252
Epoch 2/25... Discriminator Loss: 1.0833... Generator Loss: 1.0484
Epoch 2/25... Discriminator Loss: 1.2167... Generator Loss: 0.7331
Epoch 2/25... Discriminator Loss: 1.2127... Generator Loss: 1.2746

Epoch 2/25... Discriminator Loss: 1.1035... Generator Loss: 1.5120
Epoch 2/25... Discriminator Loss: 1.0620... Generator Loss: 0.9914
Epoch 2/25... Discriminator Loss: 1.2190... Generator Loss: 0.6940
Epoch 2/25... Discriminator Loss: 1.1297... Generator Loss: 1.3663
Epoch 2/25... Discriminator Loss: 1.2099... Generator Loss: 0.7329
Epoch 2/25... Discriminator Loss: 1.1160... Generator Loss: 1.4883
Epoch 2/25... Discriminator Loss: 1.3591... Generator Loss: 0.5661
Epoch 2/25... Discriminator Loss: 1.1181... Generator Loss: 1.2509
Epoch 2/25... Discriminator Loss: 1.0251... Generator Loss: 1.6326
Epoch 2/25... Discriminator Loss: 1.3711... Generator Loss: 0.5300

Epoch 2/25... Discriminator Loss: 1.1385... Generator Loss: 0.8581
Epoch 2/25... Discriminator Loss: 1.4365... Generator Loss: 0.4897
Epoch 2/25... Discriminator Loss: 1.2031... Generator Loss: 0.6752
Epoch 2/25... Discriminator Loss: 1.1732... Generator Loss: 1.9586
Epoch 2/25... Discriminator Loss: 1.1528... Generator Loss: 0.9160
Epoch 2/25... Discriminator Loss: 0.9412... Generator Loss: 1.4390
Epoch 2/25... Discriminator Loss: 1.0093... Generator Loss: 1.3092
Epoch 2/25... Discriminator Loss: 0.9700... Generator Loss: 1.0940
Epoch 2/25... Discriminator Loss: 1.0714... Generator Loss: 1.7814
Epoch 2/25... Discriminator Loss: 1.0992... Generator Loss: 1.0184

Epoch 2/25... Discriminator Loss: 1.0075... Generator Loss: 1.0427
Epoch 2/25... Discriminator Loss: 1.0285... Generator Loss: 0.9092
Epoch 2/25... Discriminator Loss: 0.9919... Generator Loss: 1.0703
Epoch 3/25... Discriminator Loss: 1.3380... Generator Loss: 0.5717
Epoch 3/25... Discriminator Loss: 0.8621... Generator Loss: 1.2369
Epoch 3/25... Discriminator Loss: 3.8989... Generator Loss: 0.0716
Epoch 3/25... Discriminator Loss: 1.0945... Generator Loss: 0.9573
Epoch 3/25... Discriminator Loss: 0.9904... Generator Loss: 1.1427
Epoch 3/25... Discriminator Loss: 1.0965... Generator Loss: 0.8691
Epoch 3/25... Discriminator Loss: 1.1366... Generator Loss: 0.8186

CelebA¶

Run your GANs on CelebA. It will take around 20 minutes on the average GPU to run one epoch. You can run the whole epoch or stop when it starts to generate realistic faces.

batch_size = 128
z_dim = 512
learning_rate = 0.0002
beta1 = 0.5


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 10

celeba_dataset = helper.Dataset('celeba', glob(os.path.join(data_dir, 'img_align_celeba/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, celeba_dataset.get_batches,
          celeba_dataset.shape, celeba_dataset.image_mode)

Epoch 1/10... Discriminator Loss: 0.6201... Generator Loss: 2.8750
Epoch 1/10... Discriminator Loss: 0.5115... Generator Loss: 3.1992
Epoch 1/10... Discriminator Loss: 0.5446... Generator Loss: 3.0213
Epoch 1/10... Discriminator Loss: 0.5487... Generator Loss: 6.8241
Epoch 1/10... Discriminator Loss: 0.6515... Generator Loss: 2.0392
Epoch 1/10... Discriminator Loss: 0.4798... Generator Loss: 2.9571
Epoch 1/10... Discriminator Loss: 0.5497... Generator Loss: 2.1911
Epoch 1/10... Discriminator Loss: 0.6562... Generator Loss: 2.0632
Epoch 1/10... Discriminator Loss: 0.7137... Generator Loss: 1.5585
Epoch 1/10... Discriminator Loss: 0.6786... Generator Loss: 2.4273

Epoch 1/10... Discriminator Loss: 1.0594... Generator Loss: 1.0628
Epoch 1/10... Discriminator Loss: 1.0706... Generator Loss: 0.9463
Epoch 1/10... Discriminator Loss: 0.9153... Generator Loss: 1.7193
Epoch 1/10... Discriminator Loss: 0.7754... Generator Loss: 1.5922
Epoch 1/10... Discriminator Loss: 1.1161... Generator Loss: 0.8112
Epoch 1/10... Discriminator Loss: 0.6839... Generator Loss: 2.0354
Epoch 1/10... Discriminator Loss: 1.1087... Generator Loss: 0.9201
Epoch 1/10... Discriminator Loss: 0.8029... Generator Loss: 2.4073
Epoch 1/10... Discriminator Loss: 0.5793... Generator Loss: 2.5054
Epoch 1/10... Discriminator Loss: 0.6621... Generator Loss: 3.8161

Epoch 1/10... Discriminator Loss: 0.5514... Generator Loss: 2.5901
Epoch 1/10... Discriminator Loss: 0.9320... Generator Loss: 1.3693
Epoch 1/10... Discriminator Loss: 1.2906... Generator Loss: 0.6721
Epoch 1/10... Discriminator Loss: 0.6146... Generator Loss: 1.9471
Epoch 1/10... Discriminator Loss: 0.7180... Generator Loss: 2.3303
Epoch 1/10... Discriminator Loss: 0.6131... Generator Loss: 1.8003
Epoch 1/10... Discriminator Loss: 0.5110... Generator Loss: 2.5356
Epoch 1/10... Discriminator Loss: 0.7215... Generator Loss: 1.4163
Epoch 1/10... Discriminator Loss: 0.9729... Generator Loss: 2.5961
Epoch 1/10... Discriminator Loss: 1.0155... Generator Loss: 4.0743

Epoch 1/10... Discriminator Loss: 0.7812... Generator Loss: 1.5927
Epoch 1/10... Discriminator Loss: 1.0552... Generator Loss: 0.9819
Epoch 1/10... Discriminator Loss: 1.5931... Generator Loss: 4.8741
Epoch 1/10... Discriminator Loss: 0.8742... Generator Loss: 1.9016
Epoch 1/10... Discriminator Loss: 1.2488... Generator Loss: 0.6981
Epoch 1/10... Discriminator Loss: 0.8239... Generator Loss: 1.4557
Epoch 1/10... Discriminator Loss: 0.5997... Generator Loss: 2.3956
Epoch 1/10... Discriminator Loss: 0.6405... Generator Loss: 2.7837
Epoch 1/10... Discriminator Loss: 0.6586... Generator Loss: 1.6713
Epoch 1/10... Discriminator Loss: 0.8776... Generator Loss: 2.2224

Epoch 1/10... Discriminator Loss: 0.6282... Generator Loss: 2.4382
Epoch 1/10... Discriminator Loss: 0.6455... Generator Loss: 2.0032
Epoch 1/10... Discriminator Loss: 0.9564... Generator Loss: 1.1896
Epoch 1/10... Discriminator Loss: 0.6090... Generator Loss: 2.4216
Epoch 1/10... Discriminator Loss: 1.0750... Generator Loss: 0.8788
Epoch 1/10... Discriminator Loss: 1.6895... Generator Loss: 4.9181
Epoch 1/10... Discriminator Loss: 0.9225... Generator Loss: 1.7197
Epoch 1/10... Discriminator Loss: 0.8492... Generator Loss: 1.3033
Epoch 1/10... Discriminator Loss: 0.9295... Generator Loss: 2.1103
Epoch 1/10... Discriminator Loss: 1.1004... Generator Loss: 0.8364

Epoch 1/10... Discriminator Loss: 0.9342... Generator Loss: 2.8947
Epoch 1/10... Discriminator Loss: 0.7375... Generator Loss: 1.6109
Epoch 1/10... Discriminator Loss: 0.8065... Generator Loss: 1.8005
Epoch 1/10... Discriminator Loss: 1.3009... Generator Loss: 0.6991
Epoch 1/10... Discriminator Loss: 1.0821... Generator Loss: 1.0815
Epoch 1/10... Discriminator Loss: 0.8731... Generator Loss: 1.2169
Epoch 1/10... Discriminator Loss: 1.1837... Generator Loss: 0.9333
Epoch 1/10... Discriminator Loss: 0.9418... Generator Loss: 3.3161
Epoch 1/10... Discriminator Loss: 1.2185... Generator Loss: 0.7113
Epoch 1/10... Discriminator Loss: 0.9303... Generator Loss: 1.1518

Epoch 1/10... Discriminator Loss: 0.9854... Generator Loss: 1.3254
Epoch 1/10... Discriminator Loss: 0.8923... Generator Loss: 1.2763
Epoch 1/10... Discriminator Loss: 1.1018... Generator Loss: 0.9109
Epoch 1/10... Discriminator Loss: 1.2617... Generator Loss: 0.6517
Epoch 1/10... Discriminator Loss: 0.8198... Generator Loss: 1.8251
Epoch 1/10... Discriminator Loss: 0.9399... Generator Loss: 1.4217
Epoch 1/10... Discriminator Loss: 0.6023... Generator Loss: 1.9847
Epoch 1/10... Discriminator Loss: 0.9608... Generator Loss: 1.3243
Epoch 1/10... Discriminator Loss: 1.0190... Generator Loss: 1.1953
Epoch 1/10... Discriminator Loss: 0.9994... Generator Loss: 1.6198

Epoch 1/10... Discriminator Loss: 0.9578... Generator Loss: 1.0649
Epoch 1/10... Discriminator Loss: 1.2123... Generator Loss: 0.7038
Epoch 1/10... Discriminator Loss: 0.5669... Generator Loss: 2.3385
Epoch 1/10... Discriminator Loss: 0.8684... Generator Loss: 1.4320
Epoch 1/10... Discriminator Loss: 1.1251... Generator Loss: 0.9583
Epoch 1/10... Discriminator Loss: 1.0672... Generator Loss: 0.9620
Epoch 1/10... Discriminator Loss: 1.7385... Generator Loss: 0.3717
Epoch 1/10... Discriminator Loss: 0.6982... Generator Loss: 1.9603
Epoch 1/10... Discriminator Loss: 1.2804... Generator Loss: 0.6736
Epoch 1/10... Discriminator Loss: 0.8923... Generator Loss: 1.2521

Epoch 1/10... Discriminator Loss: 1.0849... Generator Loss: 1.0420
Epoch 1/10... Discriminator Loss: 0.9569... Generator Loss: 1.2839
Epoch 1/10... Discriminator Loss: 1.0775... Generator Loss: 2.0768
Epoch 1/10... Discriminator Loss: 1.0123... Generator Loss: 1.0546
Epoch 1/10... Discriminator Loss: 1.1000... Generator Loss: 0.8592
Epoch 1/10... Discriminator Loss: 1.0109... Generator Loss: 0.9690
Epoch 1/10... Discriminator Loss: 0.7181... Generator Loss: 2.7212
Epoch 1/10... Discriminator Loss: 1.0038... Generator Loss: 1.1802
Epoch 1/10... Discriminator Loss: 1.2263... Generator Loss: 0.6771
Epoch 1/10... Discriminator Loss: 1.4007... Generator Loss: 0.5496

Epoch 1/10... Discriminator Loss: 1.0365... Generator Loss: 1.0056
Epoch 1/10... Discriminator Loss: 1.1941... Generator Loss: 2.6138
Epoch 1/10... Discriminator Loss: 0.9958... Generator Loss: 2.1057
Epoch 1/10... Discriminator Loss: 1.1492... Generator Loss: 0.7483
Epoch 1/10... Discriminator Loss: 1.1695... Generator Loss: 0.6936
Epoch 1/10... Discriminator Loss: 0.9871... Generator Loss: 1.8547
Epoch 1/10... Discriminator Loss: 1.0306... Generator Loss: 0.9835
Epoch 1/10... Discriminator Loss: 1.0245... Generator Loss: 0.8465
Epoch 1/10... Discriminator Loss: 0.8188... Generator Loss: 1.5721
Epoch 1/10... Discriminator Loss: 0.6964... Generator Loss: 1.8643

Submitting This Project¶

When submitting this project, make sure to run all the cells before saving the notebook. Save the notebook file as "dlnd_face_generation.ipynb" and save it as a HTML file under "File" -> "Download as". Include the "helper.py" and "problem_unittests.py" files in your submission.